ABSTRACTS
EDUCATION SESSIONS
— TUESDAY
ES23.01
Mediastinoscopy with Invasive Staging: Are They Still
Crucial?
A. Turna Department of Thoracic Surgery, Istanbul University-Cerrahpas¸a Cerrahpasa Medical School, Istanbul/TR
There are more multiple treatment strategies for non-small cell lung cancer (NSCLC) that should be selected based on staging of the disease. Nodal status indicates N component of staging and studies invariably show that upfront surgical resection of patients with mediastinal lymph node involvement (i.e., N2 or N3) is not recommended. Evidence sug-gest that, T1-4N2-3M0-1c patients shouldfirstly receive oncological treatment; otherwise, surgical treatment could be deemed to be futile. Accordingly, mediastinal lymph node involvement prediction as accu-rate as possible is recommended before any treatment planning. PET-CT, Endobronchial ultrasonography-transbronchial aspiration (EBUS-TBNA), endoscopic ultrasound-guided-fine-needle aspiration (EUS-FNB) have all inherent limitations. Mediastinoscopy has been defined to be gold standard for preoperative disclosure of mediastinal lymph node metastasis. However, small biopsy material due to being an incisional biopsy, practically low number of explored mediastinal stations (usu-ally median number of 2 or 3) led to approximately 10% of false negativity rate. Also, studies showed that, even the fact that, media-stinoscopy has been recommended to be performed in all patients except the patients with non-discrete lymph node involvement or in the patients with peripheral cT1a-cN0M0 patients, a fraction of thoracic surgeons prefer to comply with the published guidelines. Video-assis-ted mediastinoscopy lymphadenectomy(VAMLA) was developed to reduce the false negativity rate below statistically non-significant levels (below 5%). It involves dissection of at least 5 lymph nodes stations and some evidence suggest that, VAMLA is associated with better survival rate beyond selection bias phenomenon. Transcervical extended mediastinal lymphadenectomy (TEMLA) is a technically more advanced mediastinal lymph node dissection procedure that is a de fi-nition of a resection of lymph nodes from #1-9 bilaterally including aorticopulmonary and anterior mediastinal lymph nodes. The accuracy of TEMLA has been reported to be 98.4%. Taking all those achieve-ments into consideration, VAMLA or TEMLA or at least video-media-stinoscopy should be performed before selecting a therapeutical option in a patient with potentially resectable operable NSCLC. However, recent advancements in computational science could propose us that, possibly, there is enough information for us to predict mediastinal lymph node positivity without performing any invasive procedure. Artificial Intelligence (AI) is accomplished by computers that use al-gorithms, pattern matching, rules, deep learning and cognitive computing to approximate conclusions using previously defined analog or digital parameters. AI aimed to mimic the brain’s neural networks. It uses multiple layers of non-linear processing units to teach itself how to understand data classifying the record or making predictions. In a study, we aimed to evaluate the value of artificial neural network (ANN) for mediastinal nodal metastasis, by using only clinical and radiologic data. In our data set, ANN predicted mediastinal nodal involvement perfectly (AUC:1) in both training and test groups. When we used‘traditional’ univariate and multivariate analyses, younger age (<65) (AUC:0.59) and higher SUVmax (>2.5) (AUC:0.67) were associ-ated to be mediastinal nodal involvement. ANN prediction was better and it was even more sensitive than VAMLA! However, specificity of ANN resulted to be less than 0.9 in some training analyses. The major limitations of ANN include its variability, transparency and non-consistency. Nevertheless, there is a possibility that, ANN could provide
better predictions and it may help us to identify and narrow down the patients who need invasive staging. However, the usage of ANN in medicine has been continuously expanding. Future studies are needed to understand the exact place of ANN in mediastinal staging. Keywords: Non-small cell lung cancer, MEDIASTINAL STAGING, Arti-ficial Intelligence
ES23.02
Which N2 Patients Are Candidates to Surgery in the
Era If I/O?
E. Vallieres Thoracic Surgery, Swedish Cancer Institute, Seattle, WI/US The role of surgery in patients with pre-resection documented N2 disease remains a subject of controversy. In some institutions, any clinical N2 disease identified preoperatively is considered non-surgical and these patients are offered upfront definitive chemoradiation ther-apy (DefCRT) without planned resection. In other institutions, a se-lective approach to N2 patients will consider surgery as part of a multimodality approach where surgery may be offeredfirst followed by adjuvant cytotoxic chemotherapy (AC) with or without sequential post-operative radiation therapy (PORT), or where surgery will be offered after induction chemotherapy or induction chemoradiation therapy. Due to variability in N2 disease presentation, factors determining this selective approach vary between institutions and may include the bulk of N2 nodal involvement (size), extent of N2 involvement (single vs multistation, microscopic vs macroscopic), the presence or absence of extracapsular nodal involvement, the need to perform a pneumonec-tomy or not, and mediastinal sterilization after induction therapy. Historically, response to induction therapy with clearance of N2 nodal involvement following induction therapy has been shown to bode for a better prognosis after surgical resection. Unfortunately, even after complete resection following induction therapy, many patients develop distant metastases, with brain metastases prevailing. In 2018, the PA-CIFIC trial showed that adding an immune checkpoint inhibitor (IO) durvalumab for up to one year after completion of DefCRT in non-surgical stages cIIIa and cIIIB patients led to a significant and un-precedented overall survival in this population of patients. (Antonia, NEJM 2018) In patients with resectable NSCLC (cIIIA and less), recent small phase I clinical trials of either single-agent induction IO (Forde, NEJM 2018) or concurrent induction chemoimmunotherapy (Pro-vencio, JTO abst. 2018) given before surgery have demonstrated feasibility, acceptable toxicity, and unprecedented pathological response rates. It remains to be seen if these pathologic responses will translate into improved overall survival in this patient population. Extrapolating from these early observations, one may think that in-duction IO (likely with concurrent cytotoxic chemotherapy) may possibly allow us to offer surgery to a larger proportion of patients with clinical N2 disease in the future as we observe higher response rates to induction therapies which may translate in better survival. Others may want to extrapolate from the Pacific trial results and hy-pothesize that surgery followed by adjuvant IO may become a desirable option. Though encouraging, there is a paucity of data to help guide us in the incorporation of IO therapy perioperatively for patients with clinical N2 disease. As such, the role for IO in multimodality treatment for N2 disease remains undefined. Unknown are the true impacts of periop IOs in this patient population and what is the optimal combi-nation and timing of these multimodality treatments. There are more unanswered questions than established guidelines: (1) whether“IO first then surgery” is superior to “surgery first then adjuvant IO”, (2) whether an induction IO strategy followed by resection would be su-perior to DefCRT followed by IO in this population of potentially
